← Back to Kontasks

Quick Turn: Prompt Caching Implemented

Added cache_control header to Quick Turn for ~90% cost reduction on cached system prompts

✓ Completed
on 5 Jan 2026, 2:26 pm

Output

✅ Quick Turn Prompt Caching Implemented

Committed 53bf53e — pushed to main

What Changed

providers/index.ts Added cache_control header + anthropic-beta header
quick-turn.ts Added stable default system prompt for cache hits

How It Works

// System prompt with cache_control
{
  "system": [{
    "type": "text",
    "text": "You are a helpful AI assistant...",
    "cache_control": { "type": "ephemeral" }
  }]
}
  • Cache TTL: 5 minutes on Anthropic's side
  • Cache Key: Exact byte-for-byte prefix match
  • Savings: ~90% cost reduction on cached portion

Default System Prompt

You are a helpful AI assistant in Quick Turn mode.
Quick Turn is optimized for fast, concise answers without tool access.

Guidelines:
- Be concise and direct
- Provide accurate information  
- If you don't know something, say so
- Format responses clearly (use markdown when helpful)
- For complex questions, suggest using the full Claude session

Expected Savings

Scenario Before After Savings
First request (cold) $0.003 $0.003 0%
Subsequent requests $0.003 $0.0005 ~83%
10 requests in 5 min $0.030 $0.0075 ~75%

What's Next?

Restart konsole to apply changes, then test Quick Turn.

Quick Actions

Original Request

No layout configured

Details

Type General
Status Completed
Scope vibetools
Tags quick-turncachingperformancecost
Created 5 Jan 2026, 2:26 pm
Updated 5 Jan 2026, 2:26 pm
Timing
⏱️ 112.7s
14 tools
● TTFE 5% ● Thinking 81% ● Tools 14%
Created By claude

Raw Data

{
  "id": "5e8f613a-7584-4b6c-9bee-86e444ad43de",
  "type": "general",
  "status": "completed",
  "title": "Quick Turn: Prompt Caching Implemented",
  "description": "Added cache_control header to Quick Turn for ~90% cost reduction on cached system prompts",
  "context": {
    "output": "<div style=\"background:#1e1e1e;color:#e0e0e0;padding:20px;border-radius:8px;font-family:'Consolas','Monaco',monospace;\">\n<h2 style=\"font-family:sans-serif;color:#fff;margin:0 0 16px 0;\">✅ Quick Turn Prompt Caching Implemented</h2>\n\n<p style=\"color:#9ca3af;font-family:sans-serif;margin:0 0 16px 0;\">Committed <code style=\"background:#2d2d2d;padding:2px 6px;border-radius:4px;\">53bf53e</code> — pushed to main</p>\n\n<div style=\"background:#2d2d2d;padding:16px;border-radius:6px;margin-bottom:16px;\">\n<h3 style=\"color:#e5c07b;margin:0 0 12px 0;font-family:sans-serif;\">What Changed</h3>\n\n<table style=\"width:100%;border-collapse:collapse;font-size:0.9rem;\">\n<tr style=\"border-bottom:1px solid #444;\">\n<td style=\"padding:8px 0;color:#61afef;width:150px;\">providers/index.ts</td>\n<td style=\"padding:8px 0;color:#e0e0e0;\">Added <code style=\"background:#1a1a1a;padding:2px 4px;border-radius:2px;\">cache_control</code> header + <code style=\"background:#1a1a1a;padding:2px 4px;border-radius:2px;\">anthropic-beta</code> header</td>\n</tr>\n<tr>\n<td style=\"padding:8px 0;color:#61afef;\">quick-turn.ts</td>\n<td style=\"padding:8px 0;color:#e0e0e0;\">Added stable default system prompt for cache hits</td>\n</tr>\n</table>\n</div>\n\n<div style=\"background:#2d2d2d;padding:16px;border-radius:6px;margin-bottom:16px;\">\n<h3 style=\"color:#e5c07b;margin:0 0 12px 0;font-family:sans-serif;\">How It Works</h3>\n\n<pre style=\"background:#1a1a1a;padding:12px;border-radius:4px;overflow-x:auto;font-size:0.85rem;color:#abb2bf;margin:0 0 12px 0;\"><span style=\"color:#7f848e;\">// System prompt with cache_control</span>\n{\n  <span style=\"color:#e06c75;\">\"system\"</span>: [{\n    <span style=\"color:#e06c75;\">\"type\"</span>: <span style=\"color:#98c379;\">\"text\"</span>,\n    <span style=\"color:#e06c75;\">\"text\"</span>: <span style=\"color:#98c379;\">\"You are a helpful AI assistant...\"</span>,\n    <span style=\"color:#e06c75;\">\"cache_control\"</span>: { <span style=\"color:#e06c75;\">\"type\"</span>: <span style=\"color:#98c379;\">\"ephemeral\"</span> }\n  }]\n}</pre>\n\n<ul style=\"margin:0;padding-left:20px;line-height:1.8;color:#e0e0e0;\">\n<li><strong style=\"color:#98c379;\">Cache TTL:</strong> 5 minutes on Anthropic's side</li>\n<li><strong style=\"color:#98c379;\">Cache Key:</strong> Exact byte-for-byte prefix match</li>\n<li><strong style=\"color:#98c379;\">Savings:</strong> ~90% cost reduction on cached portion</li>\n</ul>\n</div>\n\n<div style=\"background:#2d2d2d;padding:16px;border-radius:6px;margin-bottom:16px;\">\n<h3 style=\"color:#e5c07b;margin:0 0 12px 0;font-family:sans-serif;\">Default System Prompt</h3>\n\n<pre style=\"background:#1a1a1a;padding:12px;border-radius:4px;overflow-x:auto;font-size:0.85rem;color:#98c379;margin:0;\">You are a helpful AI assistant in Quick Turn mode.\nQuick Turn is optimized for fast, concise answers without tool access.\n\nGuidelines:\n- Be concise and direct\n- Provide accurate information  \n- If you don't know something, say so\n- Format responses clearly (use markdown when helpful)\n- For complex questions, suggest using the full Claude session</pre>\n</div>\n\n<div style=\"background:#2d2d2d;padding:16px;border-radius:6px;margin-bottom:16px;\">\n<h3 style=\"color:#e5c07b;margin:0 0 12px 0;font-family:sans-serif;\">Expected Savings</h3>\n\n<table style=\"width:100%;border-collapse:collapse;font-size:0.9rem;\">\n<tr style=\"background:#1a1a1a;\">\n<th style=\"padding:8px;text-align:left;color:#fff;\">Scenario</th>\n<th style=\"padding:8px;text-align:left;color:#fff;\">Before</th>\n<th style=\"padding:8px;text-align:left;color:#fff;\">After</th>\n<th style=\"padding:8px;text-align:left;color:#fff;\">Savings</th>\n</tr>\n<tr style=\"border-bottom:1px solid #444;\">\n<td style=\"padding:8px;color:#e0e0e0;\">First request (cold)</td>\n<td style=\"padding:8px;color:#e06c75;\">$0.003</td>\n<td style=\"padding:8px;color:#e06c75;\">$0.003</td>\n<td style=\"padding:8px;color:#7f848e;\">0%</td>\n</tr>\n<tr style=\"border-bottom:1px solid #444;\">\n<td style=\"padding:8px;color:#e0e0e0;\">Subsequent requests</td>\n<td style=\"padding:8px;color:#e06c75;\">$0.003</td>\n<td style=\"padding:8px;color:#98c379;\">$0.0005</td>\n<td style=\"padding:8px;color:#98c379;font-weight:bold;\">~83%</td>\n</tr>\n<tr>\n<td style=\"padding:8px;color:#e0e0e0;\">10 requests in 5 min</td>\n<td style=\"padding:8px;color:#e06c75;\">$0.030</td>\n<td style=\"padding:8px;color:#98c379;\">$0.0075</td>\n<td style=\"padding:8px;color:#98c379;font-weight:bold;\">~75%</td>\n</tr>\n</table>\n</div>\n\n<div style=\"background:#2d2d2d;padding:16px;border-radius:6px;\">\n<h4 style=\"margin:0 0 8px 0;color:#fff;font-family:sans-serif;\">What's Next?</h4>\n<p style=\"color:#9ca3af;margin:0;font-family:sans-serif;\">Restart konsole to apply changes, then test Quick Turn.</p>\n</div>\n</div>",
    "requestedAt": "2026-01-05T03:25:00.000Z",
    "requestId": "1b88130d-f1b1-4181-825e-e3203e3117fc",
    "choices": [
      {
        "label": "Restart konsole",
        "value": "Restart konsole to apply the prompt caching changes",
        "primary": true
      },
      {
        "label": "Test Quick Turn",
        "value": "Test Quick Turn to verify caching is working"
      },
      {
        "label": "View VIBE.md",
        "value": "Show me the Quick Turn section in VIBE.md"
      }
    ],
    "turnTiming": {
      "totalMs": 112689,
      "ttfeMs": 5382,
      "thinkingMs": 91584,
      "toolExecutionMs": 15722,
      "toolCallCount": 14,
      "thinkingPct": 81,
      "toolsPct": 14,
      "ttfePct": 5
    }
  },
  "createdBy": "claude",
  "createdAt": "2026-01-05T04:26:38.561Z",
  "updatedAt": "2026-01-05T04:26:43.480Z",
  "requestId": "1b88130d-f1b1-4181-825e-e3203e3117fc",
  "scope": "vibetools",
  "tags": [
    "quick-turn",
    "caching",
    "performance",
    "cost"
  ],
  "targetUser": "claude"
}
DashboardReportsKontasksSessionsTelemetryLogs + Go